dumux issueshttps://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues2019-06-17T16:56:36Zhttps://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues/726More tools for regression tests2019-06-17T16:56:36ZTimo Kochtimo.koch@iws.uni-stuttgart.deMore tools for regression tests<!--
This form is for feature requests ONLY!
If you're looking for help check out the [readme](/README.md).
-->
**Feature request**
New ideas for regression tests. It should be possible to ignore changes at let's say a saturation front, maybe by allowing a certain number of dofs to have a error higher tolerance. Another idea would be to map the solution of an adaptive grid to some reference grid and compare the solution there (we have some L2-projection capabilities now (only 2d for non-matching grids)).
**What does this feature / why does DuMux need it**:
It would make it possible to create more robust tests in cases where this is desired / physically-sensible.https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues/725[Navier Stokes] Upwinding of transporting velocity2019-06-17T14:48:00ZMelanie Lipp[Navier Stokes] Upwinding of transporting velocityWe at some point also discussed if we wand to do upwinding also for the transporting velocity. I think it might be better to upwind both factors of the advective term. I think that would involve upwind decisions self versus opposite for the integrals along the frontal faces and upwind decisions inside versus outside velocity lateral for the integrals along the lateral faces.
For the upwind decision self or opposite, it would be based on (self+opposite)/2. For a upwind decision inside versus outside, I could imagine basing the decision on self+parallel/2.https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues/724[Navier Stokes] Gravity upwinding2019-06-17T14:48:36ZMelanie Lipp[Navier Stokes] Gravity upwindingDo we want to use the upwind-density for both half-cells instead of the currently used density of the corresponding half-cell?https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues/723[Navier Stokes] Possibly revise initial condition of instationary channel test2019-06-06T09:12:56ZMelanie Lipp[Navier Stokes] Possibly revise initial condition of instationary channel testI would say that the initial conditions of an instationary test should such that the Navier-Stokes and continuity equation can be fulfilled. Currently, we start our channel test with a velocity profile u=1 at the left boundary, u=0 everywhere else. This means that for the cells on the very left the continuity equation can not be fulfilled. @nedc had the idea to start with u=1 everywhere besides at the top and bottom boundary where we have u=0. This would be in accordance with continuity.https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues/722(Not) modifying parameter tree during simulation2019-06-04T15:11:44ZTimo Kochtimo.koch@iws.uni-stuttgart.de(Not) modifying parameter tree during simulationI think in the design of the parameter tree, it was acutally desired that the tree shouldn't be modified after the initialization (but it can be initialized several times by overwriting previous initializaitons).
However currently it is technically possible to call init anywhere (globally) and modify the params which may lead to side effects in other classes.
It's technically possible to throw an error if the tree is modified by init after getParam has been called for the first time. However in some tests (some sequential tests and the fluid system test) this bug/feature is actually used/misused.
We should decide which way we want it and think about how to enforce it backwards-compatibly.3.1https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues/721[Navier-Stokes] Revise definition of transporting velocity on boundary for mo...2019-06-06T09:31:02ZKilian Weishauptkilian.weishaupt@iws.uni-stuttgart.de[Navier-Stokes] Revise definition of transporting velocity on boundary for momentum upwindingThe transporting velocity is always
```
const Scalar transportingVelocity = faceVars.velocityLateralInside(localSubFaceIdx);
```
However, if the own scvf lies on a boundary and a tangential slip velocity is specified, we should probably rather take this value.3.1https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues/719FVElementGeometry public alias FVGridGeometry is not generic enough2019-05-31T19:19:08ZTimo Kochtimo.koch@iws.uni-stuttgart.deFVElementGeometry public alias FVGridGeometry is not generic enoughFor compatibilty with FEM this alias should be called GridGeometry. In the FEElementGeoemtry this alias is actually still missing. This is needed to implement elementSolution/evalSolution for FEM.https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues/717SPE 10 as example2019-05-29T09:00:00ZTimo Kochtimo.koch@iws.uni-stuttgart.deSPE 10 as exampleThere is some code for the SPE10 https://git.iws.uni-stuttgart.de/dumux-appl/dumux-spe10 based on some old dumux version. It would be nice to update it to the new version and implement the five-spot SPE10.https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues/714Calculate time derivatives by product rule and total derivative for compressi...2019-06-07T08:04:00ZMelanie LippCalculate time derivatives by product rule and total derivative for compressible Navier-Stokes equationFor the one-component compressible Navier-Stokes equation in connection with the ideal gas law we might do the following changes:
Momentum balance:
```math
\frac{\partial \left(\varrho v\right)}{\partial t} = \frac{\partial \left(\varrho^{n_I-1} v\right)}{\partial t}
```
Mass balance:
```math
\frac{\partial \varrho }{\partial t} =
\frac{1}{RT}\frac{\partial p}{\partial t} + \frac{-p}{RT^2}\frac{\partial T}{\partial t}
```
Energy balance:
```math
\frac{\partial \left(\varrho u^e\right) }{\partial t} = \frac{\partial \left(\varrho^{n_I-1} u^e\right) }{\partial t}
```
with $`n_I-1`$: last iteration stepMelanie LippMelanie Lipphttps://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues/713Generalize implementation of Beavers-Joseph(-Saffmann) slip condition2019-05-28T14:00:33ZKilian Weishauptkilian.weishaupt@iws.uni-stuttgart.deGeneralize implementation of Beavers-Joseph(-Saffmann) slip conditionSo far ths BJS condition only considers $`du/dy`$. We should extend this to $`(du/dy`$ + $`dv/dx)`$3.1Kilian Weishauptkilian.weishaupt@iws.uni-stuttgart.deKilian Weishauptkilian.weishaupt@iws.uni-stuttgart.dehttps://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues/711Access to problem quantities (like thermal conductivity) model dependent2019-05-28T13:54:14ZAlexander JaustAccess to problem quantities (like thermal conductivity) model dependentI observed that it is problem-/model-dependent how to obtain the thermal conductivity. This make it hard to reuse some of my functions with different problems/models.
- Would it be possible to unify the interface?
- Is there a (smart) way to generalize functions that should work for different problems/models?
## Short description
The interface to obtain model properties is not unique. Obtaining the thermal conductivity `lambda` looks different for different types of problems used.
- Using `SolidEnergy` model: `lambda = ThermalConductivityModel::effectiveThermalConductivity(volVars,problem.spatialParams(),element,fvGeometry,scv)`
- Using `NavierStokesNI` model: `lambda = volVars.effectiveThermalConductivity()`
In my case, this makes is hard to reuse an already implemented function. Moreover, it does not feel intuitive that the interface to obtain the conductivity differs (that much) between two physical models.
## Longer description
Kilian and me have been recently worked on a routine that constructs the temperature on a boundary. It basically gets a heat flux `Q` on the face of an element and the temperature `T_cc` at the cell center. Together with the thermal conductivity `lambda` and the distance from the cell center to the face `distance` obtain the temperature on the face `T_face` from the following formula:
```
Q = - lambda ( T_face - T_center ) / distance
=> T_face = - Q * distance / lambda + T_center
```
It has been implement once for a heat equation problem that inherits from `SolidEnergy`. In this case heat conductivity is obtained via `ThermalConductivityModel::effectiveThermalConductivity`. In my particular case it looks like that:
```c++
const Scalar lambda = ThermalConductivityModel::effectiveThermalConductivity(volVars,
problem.spatialParams(),
element,
fvGeometry,
scv);
```
Now, I wanted to reuse the function for a Navier-Stokes based problem that inherits `NavierStokesNI`. I tried reuse the function we had already written, but this fails since the heat conductivity has to be obtained via the `effectiveThermalConductivity` member function of an object of type `ElementVolumeVariables`. In my case it looks like:
```c++
const Scalar insideLambda = volVars.effectiveThermalConductivity();
```
Due to that I had to implement the same function twice which I would like to avoidhttps://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues/710Follow-up from "Feature/simplify effective laws"2019-05-28T09:17:25ZTimo Kochtimo.koch@iws.uni-stuttgart.deFollow-up from "Feature/simplify effective laws"The following discussion from !1605 should be addressed:
- [ ] @DennisGlaeser started a [discussion](https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/merge_requests/1605#note_26777): (+3 comments)
> I thought we removed wPhaseIdx from the Indices and made this a runtime-thing? How does this actually compile?3.1https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues/708Follow-up from "Feature/projector"2019-05-22T09:14:03ZTimo Kochtimo.koch@iws.uni-stuttgart.deFollow-up from "Feature/projector"The following discussion from !1609 should be addressed:
- [ ] @timok started a [discussion](https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/merge_requests/1609#note_26603): (+1 comment)
> You could optimize this for the different-grid case by only storing the parts of the projection that are non-zero. Then do the projection in the smaller spaces.https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues/707Generic implementation of L2-norm calculation using generic L2-projection2019-05-31T21:20:12ZTimo Kochtimo.koch@iws.uni-stuttgart.deGeneric implementation of L2-norm calculation using generic L2-projectionWith !1609 we get a generic l2-projection. In order to use it for interpolation between two arbitrary grids (arbitrary function spaces already works), we need to implement
* [x] 2d-2d intersections (!1625)
* [ ] 3d-3d intersections
That would be a great tool for convergence tests.https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues/706Add turbulent diffusion to shallow water model2019-05-21T12:20:07ZFrank PlatzekAdd turbulent diffusion to shallow water modelThe preliminary turbulent diffusion implementation in the shallow water model from Issue #657
(Turbulent diffusion in shallow water equations), is to be merged in the new structure for the shallow water model as part of the Freeflow model.Frank PlatzekFrank Platzekhttps://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues/705Additional Dirichlet constraints2019-05-20T13:47:14ZTimo Kochtimo.koch@iws.uni-stuttgart.deAdditional Dirichlet constraintsI would be nice to be able to set Dirichlet constraints for any dof in the problem.
The assembler could ask the problem if it defines additional constraints (constexpr bool func or by member function inspection (probably too magical)) and if yes implement these constraints after the assembly.
This feature could fix e.g. #704. However I'm having problems to come up with other used cases, maybe some overlapping domain decomposition stuff.https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues/704Multiphase tracer produces singular matrix if saturation is zero2019-05-28T13:31:05ZTimo Kochtimo.koch@iws.uni-stuttgart.deMultiphase tracer produces singular matrix if saturation is zeroA case that is like in two-phase models is that one phase disappears. If the tracer only exists in this
phase, the tracer also disappear, however that means that the local equation/residual degenerates.
This is automatically the case for all dofs where saturation is zero (over the whole time integration interval, see #703).
A robust solution could be to check for zero saturation and replace the equation for those dofs with the Dirichlet constraint that the concentration is zero. For that it would need to be possible to set Dirichlet for every dof even when not on the boundary, see #704.Timo Kochtimo.koch@iws.uni-stuttgart.deTimo Kochtimo.koch@iws.uni-stuttgart.dehttps://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues/703Spatial parameters cannot be time dependent if not solution-dependent2019-05-20T13:40:24ZTimo Kochtimo.koch@iws.uni-stuttgart.deSpatial parameters cannot be time dependent if not solution-dependentIf we compute the tracer model in combination with a fluid that changes properties over time, we
need the old properties to evaluate the storage.
The problem is, the volvars currently don't know anything about time, so we can't pass that information
to the spatial params. That's why we currently assume that all secondary variables are either constant
or solution-dependent but not time-dependent without being solution-dependent. This is a big drawback
in general and a bug or at least an imprecision in the tracer model.https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues/700Test virtual interface for Problem2019-05-16T15:50:49ZTimo Kochtimo.koch@iws.uni-stuttgart.deTest virtual interface for ProblemGetting rid of the problem template would be a huge improvement in the code.
We should do performance measurement and check what run-time penalty a virtual problem interface would bring.
For that we have to first check which functions actually need to be virtual, add the virtual keyword, we could then exchange templates for a base class pointer one-by-one, and check for performance penalties.
The highest penalty is probably expected for a grid with a high boundary to internal cell ratio, and simple boundary and initial conditions so that the possible virtual function call would actually matter.https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/issues/690[material][co2] create co2 tables with pressures less than 1 bar2019-05-20T07:02:22ZKatharina Heck[material][co2] create co2 tables with pressures less than 1 barthe current co2 tables start at 1 bar pressure but in the fluidsystems e.g. brineco2 we calculate e.g. the gas density based on partial pressure (which can be less than 1 bar). That can lead to wrong results if the partial pressure is less than 1 bar. Then the value of 1 bar is taken from the table and added to the density calculation (which is especially wrong if no co2 is present at all)
Possibly this could be a task for a Hiwi to look into a meaningful relationship for densities etc for co2 and create new tables.Johannes HommelJohannes Hommel